Research map / Point clouds / Segmentation and detection
ReDAL: Region-Based and Diversity-Aware Active Learning for Point Cloud Semantic Segmentation
Tsung-Han Wu, Yueh-Cheng Liu, Yu-Kai Huang, Hsin-Ying Lee, Hung-Ting Su, Ping-Chia Huang, Winston H. Hsu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use sub-scene regions instead of whole scans as query unit (ablation) | ENT selection · S3DIS Area5 | mIoU: over 10 % | Improved |
| add diversity-aware selection (cluster penalization) (ablation) | ENT + Region · S3DIS Area5 | mIoU | Improved |
| add color discontinuity and structural complexity to the score (ablation) | ReDAL · S3DIS Area5 | mIoU: helpful when labeled points larger than 9 % | Mixed |
| add color and structure to entropy without diversity module (ablation) | ENT + Region · S3DIS Area5 | mIoU | No clear effect |
| use ReDAL instead of random scan selection | SPVCNN · S3DIS Area5 (15% labeled) | mIoU: 49.02 -> 57.97 | Improved |
| use ReDAL instead of random scan selection | MinkowskiNet · SemanticKITTI (5% labeled) | mIoU: 54.67 -> 59.76 | Improved |
| use softmax entropy scan selection instead of random selection | SPVCNN · S3DIS Area5 (15% labeled) | mIoU: 49.02 -> 45.06 | Hurt |
| use Core-Set selection instead of random selection | SPVCNN · S3DIS Area5 (15% labeled) | mIoU: 49.02 -> 41.40 | Hurt |
| ReDAL labeling instead of full labels on bicycle class | ReDAL · SemanticKITTI (5% labeled) | bicycle IoU: 20.3 -> 29.5 | Improved |
| ReDAL labeling instead of full labels on average IoU | ReDAL · SemanticKITTI (5% labeled) | average IoU: 61.4 -> 59.8 | Hurt |
| vary kNN size, decay rate eta and number of clusters (ablation) | ReDAL · S3DIS / SemanticKITTI | mIoU | No clear effect |
Later papers that compared with this method
2 later accepted papers report ReDAL as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Hierarchical Point-based Active Learning for Semi-supervised Point Cloud Semantic Segmentation (ICCV 2023 · better on 1 of 1 shared result)
- LiDAL: Inter-Frame Uncertainty Based Active Learning for 3D LiDAR Semantic Segmentation (ECCV 2022 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Hierarchical Point-based Active Learning for Semi-supervised Point Cloud Semantic Segmentation (ICCV 2023 · similarity 0.88)
- ViewAL: Active Learning With Viewpoint Entropy for Semantic Segmentation (CVPR 2020 · similarity 0.88)
- Weakly Supervised Point Cloud Semantic Segmentation via Artificial Oracle (CVPR 2024 · similarity 0.86)
- LiDAL: Inter-Frame Uncertainty Based Active Learning for 3D LiDAR Semantic Segmentation (ECCV 2022 · similarity 0.86)
- Less Is More: Label Recommendation for Weakly Supervised Point Cloud Semantic Segmentation (AAAI 2024 · similarity 0.85)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
